Hi all,

I am handling a CFD problem of a complex 3D duct geometry. It has four inlets and seven outlets. The geometry has lot of joints and the ducts have sharp curvature. Also, all the inlets are at varying temperatures from 150 Deg. F to 5 Deg. F. So there is lot of mixing in the domain.

The first phase of the problem has validation with the experimental values.

i am facing lot of problems. Temperatures at the outlet deviates in the range of +2 to -9 %. I have tried almost all the ways to bring down the error.

1. The solution is already grid independent 2. y+ is in the range of 30 to 60 3. I have tried almost all the turbulence models from std K-Epsilon to RSM 4. I have already done three levels of grid adaption. 5. I have used tetrahedral mesh. (generating hexahedral mesh is very difficult)

How shall i go about. Please help me out. How much effect will the mesh type will have?. Will hexahedral mesh really help me to bring down the percent error ?. I want to know the acceptable range of percent error in a CFD problem ?.

Thanbks and Regards

I suggest you generate a boundary layer mesh using Gambit.